2
Surveying national elites in the
Federal Republic of Cermany
URSULA HOFFMANN-LANGE
Universities ofMannheim and Texas at Austin
2.1 Introduction: The West German Elite Study, 1981
Studies of national elites are concerned with the most powerful
persons in a society, i.e. persons with considerable influence on
collective decisions of central importance. This preliminary
definition of national elites will be elaborated in a subsequent
paragraph: before doing this, however, an overview ofthe theoretical
approaches of elite research, as well as a short description of the
research strategy used in the West German elite study of 1981, will be
given.
In addition to providing essential descriptive information on the
elites of a certain society, ernpirical sfudies of national elites can also
be used to test theoretical assurnptions about the relations between
elites and society. Theories of elite recruitment and elite circulation
constitute the oldest tradition in elite theory. They assurne a relationship between the character of a society, the prevailing mode of eli te
recruitrnent and of the social characteristics of elites (Bottornore,
1966). Changes in the criteria of elite recruitment and, hence, in the
social characteristics ofelites, are taken as indicators ofsocial change,
and vice versa. In this vein, it is often assurned that the transition frorn
traditional to modern industrial society has affected elite recruitrnent
by substituting achievement criteria for the forrnerly prevailing
ascriptive criteria.
Theories of conflict and consensus among elites assurne, instead, a
certain degree ofindependence of eli tes from societal restraints. They
claim that elites can reach a consensus on procedural norms, the rules
of the game, which allow peaceful conflict regulation even in
societies with deep socio-cultural cleavages (Lijphart, 1977; Field
and Higley, 1980).
28
RESEARCH ME71-IODS FOR ELITE STUDIES
The nature ofthe linkages between elites and non-elites (Putnam,
1976, eh. 6; Welsh, 1979, eh. 7; Stokes and Miller, 1962; Millerand
Stokes, 1963; Bames, 1977) in a soeiety is a third major threa~ of
theoretieal thinking about elites. lt is eoneemed with the respons1veness of elites to the demands of the general population, i.e. the
representation of interests in elite deeision-making. This can be
studied by eomparing values and issue attitudes of elites to those of
the population at large. The degree of eongruenee among different
elite and population subgroups is then used to test the adequacy of
different models of interest representation, e.g. pluralist, ruling dass,
eonsociational, power elite, or eorporatist models.
The survey approaeh in empirical elite researeh has to be
distinguished from another use of elite interviewing in which elites
serve as informants/experts about a speeific field of investigation, e.g.
Raab's study reported in Chapter 6. The two different uses of elite
interviews imply differenees in sampling and research design. While
for expert interviews a qualitative approach seems most appropriate,
quantitative methods are needed in order to gather reliable
information on backgrounds, attitudes, and activities ofelites. Critics
have often maintained that it is impossible to use such a quantitative
approaeh in elite research. They have argued that elites are reluctant
tobe interviewed by methods appropriate only for 'mass' surveys.
The faet, however, that many quantitative surveys ofnational elites
have been carried out successfully has proved them wrong.
The quantitative approach has a number of advantages as weil as
disadvantages. The use of a highly standardized questionnaire for a
broad stratum of respondents working in rather different settings
limits the depth of the information that can be collected about career
pattems, role behaviour and decision-making aetivities. Similarly,
the questions conceming perceptions of political problems and
political ideologies have to be limited to a set of forced-choiee
questions. 1
What is lost in detail, however, ean be gained in broadness. The
inclusion of different elite sectors each represented by a sufficient
number of respondents, and the imposition of a common frame of
referenee by using forced-ehoiee questions, allows study of the
pattems of dissent and consensus among different elite and
population subgroups, i.e. the structure of political cleavages in a
co~ntry. Similarly, by asking respondents for their regular interact1on partners, the overall structure of the elite network can be
a~aly~d, even when detailed information conceming the eontent,
direet1on, and frequency of these interactions is laeking.
NATIONAL EL/TES IN GERMANY
29
The West German elite study of 19812 was designed as a
quantitative, cross-sectional national elite survey. Respondents were
holders of elite positions in various sectors, i.e. political, civil service,
business, trade union, mass media, academic, military, and cultural
elites. The study is comparative in a threefold sense.
First, it allows the study of changes in the elites over time by
comparing the results to those of two previous elite surveys in West
Germany of 1968 and 1972 for which a similar design had been used
(Hoffmann-Lange et a/., 1980).
Secondly, an internationally comparative approach is ensured by
the use of a number of questions on elite networks which bad
previously been asked in the United States and Australia (Barton,
1985; Higley et a/., 1979).
Thirdly, some of the questions, mainly concerning value
orientations and issue attitudes, have also been used in a general
population survey in early 1982, thus allowing for comparisons
between elites and the population at large.
2.2 The Sampling Procedure: Methods and Theoretical
Approaches
Each sampling procedure presupposes a theoretical as weil as an
operational definition of the population about which assertions are to
be made. On the other band, most definitions of elites are rather
imprecise and give only a little guidance as to the adequate sampling
method to apply. Agreement among them is normally limited to a
common focus on the macro level of societies, institutionalized
power, and influence on collective decisions. But a definition of
national elites as 'persons with power individually, regularly, and
seriously to affect political outcomes at the macro level of organized
societies' (Higley et a/„ 1979, p. 17), still leaves a wide range of
choices to the discretion of the researcher in sampling an elite
population. lt allows for different forms of power wielding and
different power resources: direct participation in decision-making
within large-scale private and public organizations, influence on the
definition of social problems and/ or influence on public opinion.
Each of them can be legitimately considered as qualifying a person as
a member ofthe national elite.
In the reputational approach, experts are asked to indicate the
most powerfiil persons in a social system. The usefulness of this
approach is, however, limited to less complex social systems such as
small or medium-sized communities where decision-making power
30
RESEARCH METHODS FOR EUTE STUDIES
is concentrated among a readily identifiable elite group. Decisionmaking on the national level of modern societies is instead much too
complex to allow for the identification of all members of an elite by
asking only a small number of experts. Reliable, though always
subjective, knowledge about who the powerful are is usually limited
to a few decision-making arenas and to elite members themselves
since they are the ones with the most direct access to decision-making
processes. The opinions ofexperts without such a direct access to the
relevant processes are instead biased even more by subjective
preconceptions about the power structure.
The decisional approach defines power as direct participation in
political decisions. This approach has the advantage of using a
behaviourally derived measure of power, but the necessarily small
range of issues that can be studied empirically in order to identify
decision-makers makes it difficult to generalize the findings to the
entire powerstructure.
In determining .national elite samples in complex, industrial
societies, the positional approach has been the one most widely used.
lt is the easiest to apply in practice since it neither presupposes,
judges nor requires lengthy decisional studies. Starting out from a list
of elite sectors, the researcher then proceeds to select the most
important organizations within each sector. In a third step, the top
positions within each organization have to be determined. The
current incumbents of these positions are then finally considered as
members ofthe elite.
The three approaches of elite identification can be classified
according to the degree to which they allow for two dimensions of
power:
( 1) formal vs informal power;
(2) direct participation in political decision-making vs indirect
influence on political decisions.
The codified rules of political decision-making will be included in
this classification, too. Figure 2.1 shows that these latter rules use the
most restricted concept of power which the decisional as weil as the
positional approach each extend on one of the two dimensions but
~ot on the other. Finally, the reputational approach measures power
m the ~roadest s~ns~, all?wing for formal and informal power as well
as for direct and md1rect mfluence on political decisions.
_Regardless of the approach used, each operational definition of
ehtes has to solve an additional problem, namely to specify the
NA 7/0NAL ELITES IN GERMANY
31
boundaries of the elite universe, i.e. the size ofthe elite tobe studied.
Should it be limited to the very top stratum of powerfiil persons with
broad influence over a relatively wide range of decision-making
matters or should we go further down in the organizational
hierarchy? In this latter case one would also include persons with a
much more restricted range ofdecision-making power who, however,
participate more intensively in individual decisions and thus may
sometimes be even more important than those in the top stratum in
shaping these decisions.
Figure 2.1
Classification of the approaches of elite identification
Power
resources
Participation in political decision-making
Direct
participation
Direct
participation
and indirect
influence
Formal
power
Codified rules of
political
decision-making
Positional
approach
Formaland
Decisional
approach
Reputational
approach
informal
power
In the West German elite study, 1981, the positional approach was
used to define the elite universe. The positional approach was,
however, supplemented by the reputational approach: respondents
were asked to name other persons who were important for decisionmaking in their own sphere of activity. The empirical relationships
between these two approaches will be analysed in a latersection.
Starting out from a rather broad definition of positional elites,
altogether 3,580 positions in nine major elite sectors as well as a
couple of minor sectors3 were determined as belonging to the
positional sample. The criteria used for the incorporation of
positions into the sample depended on general assumptions about the
national power structure and power within and among sectors. They
were, therefore, inevitably somewhat arbitrary, and other scholars
would have come up with a partly different sample.
Experience shows, however, that disagreement concerning the
adequacy of such criteria is particularly pronounced with regard to
the sector composition and the lower boundaries of the elite sample.
The broad definition used in the West German elite ensures that at
least no important positions have been omitted. Moreover, it allows
32
RESEARCH METHODS FOR ELITE STVDIES
the study of the effects of the incl usion of certain sectors and lower
hierarchical levels on the survey results. Table 2.1 shows the sector
composition of the sample of elite positions. Due to mu_lt_iple
position-holding and transitory vacancies, the number of pos1tionholders ('target persons ') was lower than that and amounted to a total
of3,164.
Table2.1
Seetor composition of the West German e/ite study, 1981
Seetor"
Positions
%
n
Politics
Civil Service
Business
Business Associations
Trade Unions
MassMedia
Academic
Military
Cultural
Other
539
479
837
394
155
376
209
172
188
231
15. l
13.4
23.4
11.0
4.3
10.5
5.8
4.8
5.3
6.4
Position-holders
%
n
452
471
688
295
155
354
179
172
180
218
14.3
14.9
21.7
9.3
4.9
11.2
5.7
5.4
5.7
6.9
Respondents
%
n
274
296
285
174
87
222
130
43
104
129
15.7
17.0
16.3
10.0
5.0
12.7
7.5
2.5
6.0
7.4
1744 100.l
100.0
3580 100.0 3164
Total
Note: a See Appendix to this chapter for detailed Iist oforganizations and positions
2.3 Field work: Organization, Access and Problems of Data
Protection
Given the considerable size of the target population, the survey could
only be carried out in co-operation with an opinion research
institute. GETAS ofBremen, one ofthe major West German polling
institutes with sufficient experience in social research, was entrusted
with this task. lt provided the technical infrastructure, i.e. its pool of
qualified interviewers, printing services, the handling of interviewer
payments, and.the processing ofthe interviewer records.
The organization of the field work was divided between the
research team and GET AS by a margin of one-third to two-thirds.
The sample was, however, divided into two 'fields' or strata. Field 1
included the most senior position-holders for whom we expected
greater difficulties ofaccess, e.g. cabinet members, secretaries of state,
presidents of business corporations, business associations and trade
unions, editors-in-chief of the major newspapers, etc. Field II
comprised the less senior position-holders in these areas.
The two fields were then organized separately in that we used two
separate interviewer staffs: 85 interviewers in field II and 24
NATIONAL El/TES IN GERMANY 33
especially qualified interviewers in the top field. Members of the
research team belonged to the latter staff. An intensive programme of
interviewer training was also deemed necessary. This was
supplemented by a written guide containing lengthy comments about
the research goals and the intentions behind the individual questions.
At the beginning of March 1981, a personal and individually
signed letter was sent to every position-holder in the sample,
requesting an interview. The letters were posted in Mannheim in
order to document that the study was university based and not a
commercial survey. A reply postcard on which the respondents could
indicate possible interview dates accompanied each letter. In April, a
second letter (call back) was sent to those who had not responded to
the first one.
The field organization did not differ substantially between the
fields. All replies were registered by the field directors in charge. The
dates offered were checked, and appointments were confirmed either
by letter or by telephone. Refusals were mostly so definite that a
second attempt seemed unwarranted. On the other hand, the frequent
inquiries concerning the research goals and the sampling criteria
were treated with special care in order to ensure the highest response
rate possible.
Table 2.2
Response rates for successive waves of the field work
Refasals
n
%
Reactionto
first letter
First callback
Second callback
P~ition
reshuffiea
Total
n752
%52.9
n 178
% 12.5
n367
%25.8
n 123
%8.7
%
35.8
60.5
70.3
50.0
n 1420 44.9
% 100.0
n
Comp/eted
interviews
%
%
1350
77.4
116
6.7
155
8.9
123
7.1
64.2
1744
100.0
55.1
39.5
29.7
50.0
Total
n
%
%
2102
66.4
294
9.3
522
16.5
246
7.8
100.0
3164
100.0
100.0
100.0
100.0
100.0
Note: 8 Position reshuffies bad tobe treated separately because they invariably led to
a peculiar contact pattem
By mid-May, the number of replies declined sharply. Since at that
time about one-third of the selected position-holders bad not yet
reacted to either of the letters, a third wave to contact these persons
was necessary. This was done by telephone. Table 2.2 contains the
34
RESEARCH METHODS FOR ELITE 57UDIES
distribution ofresponse rates for the three waves ofthe field work. lt
shows that the second and third waves were successful with regard to
the absolute numbers ofinterviews they enabled us to conduct, even
when the response rates were much lower than for the first wave.
All contacis with target persons and their personal staff were
registered. The number of contacts required to obtain a result, either
an interview appointment or a definite refusal, is a relevant indicator
of the expenditures that are necessary to carry out a study like this.
Only those contacts were counted, however, that occurred between
the position-holders or their staff and the field directors until either a
refusal or a first interview appointment 'was reached. Date and
address were then handed over to the interviewer. The rather
frequent postponements of appointments directly arranged among
interviewer and respondents were not registered as separate contacts.
The minimum number of contacts necessary to establish a definite
outcome -was two for refusals and three for completed interviews.
Table 2.3 shows that the number of contacts increased sharply for
those persons who did not react to the letters ofthe first and second
waves. Among this group (the 'second callback'), the expenditure for
a refusal was nearly as high as that for a completed interview.
Table 2.3
Average number of contacts required to obtain a definite refusal or
an interview
Reaction to first letter
First callback
Second callback
Position reshuffie
Total
Average number ofcontacts
Total
Refusals
Completed
interviews
4.1
4.7
2.9
4.1
5.3
3.3
4.9
4.7
5.5
4.1
5.2
3.1
4.2
3.4
4.9
The field directors passed only those addresses to the interviewers
for which they had obtained the consent ofthe respondent tobe interviewed. The interviewers were asked to confirm appointment dates
and also to check the correct interview address.
The interviewer reports on the interview situation (Table 2.4)
reveal that most of the interviews were conducted under rather
favo_urabl~ circumstances. Only a few disturbances occurred during
the mterv1ews. The average interview length was 88 minutes and,
hence, somewhat shorter than the length of 90 minutes we had
announced in the letters. Differences between sectors can largely be
NATIONAL El/TES IN GERMANY 35
attributed to the familiarity of the respondents with being
interviewed and with the topics raised in the questionnaire which
were mostly political questions.
Table 2.4
Interviewer reports on interview situation
! .Evaluation ofthe interview situation
Largely without interruptions
Someinterruptions which, however, bad no influence
on the interview situation
Frequent or prolonged interruptions with negative
effects on the interview si tuation
Missing
%
n
1404
80.5
283
16.2
40
17
2.3
1.0
Good
Fairlygood
Notgood
At first good, then declining
At first bad, then improving
Missing
1459
157
36
23
37
32
83.7
9.0
2.1
3. Number ofinterruptions
None
Onetothree
More than three
Missing
975
663
84
22
55.9
38.0
4.8
4. Average length ofinterruptions
No interruptions
1-5 minutes
6-10 minutes
l l-20minutes
Morethan 20 minutes
Missing
975
456
150
93
37
33
55.9
26.1
8.6
5.3
2.1
1.9
2. Evaluation ofco-operativeness of respondents
1.3
2.1
1.8
1.3
Politicians and journalists achieved the shortest averages whereas
respondents in the business, voluntary associations, and academic
sectors were less experienced and needed more time to answer the
questions (see Table 2.5).
Due to legal regulations passed in recent years, the handling of data
protection represented a specific problem. In general population
surveys, anonymity of individual respondents in data files is usually
accomplished by separate storage of the respondents' addresses and
the survey data. Moreover, address files are normally erased
immediately after the completion ofthe field work. The anonymity
ofthe survey data is ensured by the rule that no variables are stored
36
RESEARCH METHODS FOR ELITE STUDIES
Table2.5
Response and average length of interviews
Seetor
Politics
Civil Service
Business
Business
associations
Trade unions
Massmedia
Academic
Military
Cultural
Other
Total
n
Respondents
%ofsample
total
Response
rat es
Average
lengthof
interviews
274
296
285
15.7
17.0
16.3
60.6
62.8
41.4
86.5
84.3
90.4
174
87
222
130
43
104
129
10.0
5.0
12.7
7.5
2.5
6.0
7.4
59.0
56.1
62.7
72.6
25.0
57.8
59.2
91.5
92.5
83.5
90.9
85.5
84.5
97.2
1744
100.1
55.1
88.1
that allow for identification of individual respondents either alone or
in combination with other variables, e.g. locus of residence, füll
address, employer, etc. In the case ofelite respondents, however, such
protection of the survey data is not possible because meaningful
analyses ofthe survey data presuppose additional information about
positional characteristics ofthe respondents.
The legal regulations of data protection permit the storage and
analysis of personal data of this kind only under the condition that the
respondents declare their explicit consent in written form.
Respondents were therefore asked at the outset of the intetview to
sign a special form designed for this purpose. The form also contained
information about the precautions taken to ensure confidential
handling of the survey data at the University of Mannheim. This
procedure which bad been tested in the pretest of the study did not
produce difficulties throughout the field period.
2.4 Response Rates and lmplications of the Sampling Design for
Analysis
The field period lasted from late March until the end of July, 1981.
The overall response rate was nearly the same in both fields. At 55.1
per cent it is somewhat lower than the one obtained in the 1972 elite
study and corresponds exactly to that of the 1968 elite study.
However, the analysis problems which were created by extending the
field period would have been greater than the advantages of a slightly
NATIONAL ELITES IN GERMANY 37
higher response rate since the evaluation of political questions may
be influenced by political events during the field period.
Table 2.5 contains the response rates for the different sectors.
Compared to those of the two previous surveys they, too, show a
remarkable continuity over time. Only the response rates in the
sectors for military and trade unions have declined by more than l 0
per cent since 1972. In the case of the trade unions, this is presumably
due to the fact that a number of unions were involved in wage
negotiations during the field work period. In the military sector, the
main reason lay with the Federal Ministry ofDefence which had been
asked for a special permit for the military leaders to participate in the
study long before the beginning of the field work. The permit was
granted, however, only after more than one month offield work had
already elapsed.
Altogether, 1,420 of the target persons could not be interviewed. Of
the latter, 110 bad expressed their general agreement to be interviewed but appointments could not be made due to the difficulties of
finding a free date at which the interview could take place. Reference
to an overcommitted time budget was the single most frequently
mentioned reason. However, that claim could also have been
legitimately used by any of the position-holders contacted. A tiny
minority mentioned general reservations about survey research, and
some expressed doubts concerning the confidentiality ofthe data.
Scholars have frequently suspected that the refusal rate in elite
surveys increases as high er levels in the hierarchy are reached but that
this fact is usually concealed because the relevant response rates are
not reported. In order to test this assumption the elite sample was
subdivided by seniority of position and separate response rates for
these two elite strata were calculated. This was done by using similar
classification criteria as in the assignment to the two interview fields.
Table 2.6 indicates that the response rate shows no linear and simple
relation to seniority of position. The suspected effect exists only in
the sectors for politics and business associations. In the civil service,
business, and trade union elites, response rates were instead
somewhat higher in the top stratum.
The experience of the field work did not convey any testable
suggestions conceming the factors determining individual reluctance
or willingness tobe interviewed for the study. A thorough analysis
revealed no serious distortions of the sample of respondents as
compared to the original sample. This means that the results can be
viewed as giving a fairly true portrait ofthe entire West German elite
sample with regard to the social characteristics of this group.
38 RESEARCHMETHODS FOR EUTE STUDIES
Table2.6
Response rate and seniority of position
Sectof"
Politicslb
Politicsll
Civil service 1c
Civil service II
Businessl
Business II
Business associations 1
Business associations II
T rade unions 1
T rade unions II
Massmedial
Mass media II
Positionho/ders
n
246
206
163
308
242
446
61
235
33
122
88
266
Comp/eted
interviews
n
133
141
114
182
116
169
29
145
19
68
51
165
Response
rate
%
54.l
68.4
69.9
59.l
47.9
37.9
47.5
61.7
57.6
55.1
64.8
62.0
Notes:
For the academic and cultural elites which do not display a comparably clear
organizational hierarchy, no such subdivision was tested
6 In this group, the especially low response rate among the members ofthe Federal
Govemment is compensated by a rather satisfactory one among those of state
govemments
.
.
c Deviating from the classification for the field work, all Secretanes of State m
Federal and State Ministries were counted as belonging to the top stratum
a
Moreover, given the predominantly conservative political
preferences of the respondents, the danger that they represent only
the more liberal part ofthe West German eli te can be ruled out.
Elite sampfing inevitably produces weighting problems due to
power differences within an elite. Unlike voting where each vote
counts the same, unequal influence has to be assumed in collective
decision-making. The results of unweighted analyses are instead
affected by the sample composition chosen by the researcher with
regard to the inclusion of sectors, organizations, and positions.
Predictions ofdecision-making outputs on the basis ofan unweighted
analysis of elite attitudes can thus be highly misleading, particularly
when differences of opinion exist within an elite, e.g. between sectors
or competing parties. lt is necessary to keep different subgroups apart
and to avoid unwarranted aggregations. lnferences about 'the elite'
should be made with care and only after having analysed subgroups
separately. Table 2. 7 shows how much the results for a number ofkey
variables differbetween sectors. Furthermore, the subdivision ofthe
main sectors according to positional subgroups shows to what extent
the inclusion of second level position-holders affects the results for
NA TJONAL ELITES IN GERMANY 39
the selected variables. Though this effect is not very pronounced for
many variables, it should nevertheless not be considered as
negligible.
2.5 A New Approach for Locating National Elites
Sole reliance on the positional approach yields a sample of elite
position-holders and allows the study of formal power. Although
most scholars agree that formal competence derived from
incumbency ofleadership positions plays a much more crucial role in
modern societies than in simpler systems, an identity in the
structuring of formal and informal influence cannot be assumed.
This means that power is never perfectly correlated with position
(Putnam, 1976, p. 16; Scheuch, 1973, p. 1005 ff.). Ifwe are interested
in making inferences not only about the sample of position-holders
but also about the group of persons actually most influential in
national decision-making, a weight for the actual influence of
different persons has to be found. This will solve two fundamental
and interrelated problems of the positional approach simultaneously, namely differences in influence and the boundary
problem.
Differences in influence can be caused by differences in formal
competence as well as by the varying degrees to which formal
competence is being transformed into actual influence by a person
(cf. Mokken and Stokman, 1976, p. 52 f.). The boundary problem is
likewise a twofold one: the positional approach does not provide a
single criterion for determining the boundaries in different sectors
and subgroups. The sector composition ofa positionally defined elite
sample reflects instead a priori resources, e.g. political decisionmaking authority, economic power, influence on public opinion, etc.
In order to compare the influence or power of different persons or
subgroups and to determine the overall boundaries of an elite
population, an empirical measure of influence and a uniform
boundary criterion are needed.
Since the late 1960s, various sociometric methods have been
proposed that allow the empirical study ofinfluence relations among
elites (e.g. Kadushin, 1968; Laumann and Pappi, 1976; Moore, 1979;
Higley and Moore, 1981). This is generally done by asking a
positionally defined sample of elite respondents for interaction
partners. Such an approach was also used in the West German elite
study of 1981. The procedure chosen followed closely Kadushin's
theoretical concept of social circles and its operationalization in two
75.0
28.3
7.1
72.3
44.3
43.8
30.0
43.8
48.0
53.7
51.0
54.4
Business(n=459)
54.8
A Chairmen ofboards
in business enterprises and
business associations
(n= 184)
55.7
B Other members of
boards in business and
business associations
(n=238)
44.7
7.2
75.0
37.8
53.l
Civil Service(n=296)
A Secretaries ofState
and Dept. Heads
Federal Ministries
(politically appointed)
(n=65)
B Secretaries ofState
Ministries (n = 100)
C Other (permanent
subdept. heads in
Fed. Ministries;
heads ofEC, Federal
and State Agencies
(n=l3l)
53.8
74.0
32.4
7.4
72.3
20.9
93.9
94.9
92.2
93.9
60.0
5.6
4.9
4.7
4.3
5.2
31.8
53.8
87.9
64.6
67.7
100.0
100.0
3.9
49.0
76.5
74.4
45.9
42.4
26.2
40.4
47.3
48.l
47.5
16.4
7.8
6.6
6.5
19.2
6.8
5.5
5.3
5.2
5.4
5.0
5.2
5.0
6.5
28.7
14. l
24.6
22.7
11.4
7.4
10.2
17.7
7.2
• 24.6
43.4
49.2
36.5
40.8
44.4
1
41.9
5.5
33.8
100.0
4.2
17.8
50.5
64.9
28.9
24.7
51.3
49.4
Politics(n=274)
A Cabinet members
(Federal and State
Governments)(n=82)
B Other politicians
(n= 192)
52.3
6.4
35.7
52.7
Total (n= 1744)
68.5
Party preference
Mean o/o Upper and
Mean %Party
Left-right
o/o
age upper rniddle Acadernic years in rnembers CDU/CSU SPD
FDP scale(mean)
degree position
dass origin
Seetor>
55.0
59.4
48.2
Military(n=43)
Judiciary(n=38)
Cultura/ elite(n= 104)
Other(n=91)
Notes:
48.4
28.9
37.8
53.7
59.3
61.5
100.0
32.6
96.2
7.1
7.7
5.3
2.1
5.9
7.5
41.3
39.1
51.4
46.6
35.6
57.8
21.4
65.8
14.6
20.0
47.2
52.9
74.4
50.0
55.9
29.2
18.6
49.5
13.6
11.5
12.9
73.0
33.5
100.0
96.3
98.9
5.4
39.3
41.0
26.5
7.7
23.7
26.0
19.1
23.8
83.1
84.6
83.5
6.7
27.0
17.6
15.4
25.4
23.3
25.0
23.8
1.7
3.8
2.4
18.9
6.5
5.2
4.3
5.2
6.7
5.5
5.2
5.6
5.3
4.2
4.0
4.1
a Results for subsectors deviating for a response category by more than 10 percentage points from the percentage of the whole sector (for
percentage-based indicators), or by more than two units from the sector mean (for indicators expressed as means), are printed in italic.
b The assignment ofrespondents to the subgroups in the analysis deviates slightly from the classification used in previous tables.
54.4
51.3
24.6
50.6
(n= l 50)
Science(n= 130)
9.1
59.7
54.6
8.8
8.5
43.8
30.7
51.9
Mass Media(n=222)
6.7
9.5
8.7
8.0
7.4
53.4
11.1
8.0
59.5
47.3
8.7
55.1
A Chairmen and vicechairmen (n=27)
B Other members of
boards, heads of
districts (n = 60)
A Chiefeditorsof
major newspapers,
directors of
broadcasting stations
(n=72)
B Other leading
joumalists in the press
andin broadcasting
7.8
Trade Unions(n=81)
17.l
56.4
54.0
C Agricultural
associations (n=37)
42
RESEARCH METHODS FOR EUTE STUDIES
previous studies of national elites in the U nited States (1971172) and
Australia (1975). Respondents were first asked to indicate the one
national issue on which they had been most active during the last
year. After having described the nature of this issue in some detail,
they were further asked to name their most important interaction
partners in the context ofthis issue.
This question can be regarded as measuring instrumental
reputation for political influence. Since the focus was on issues of
more than intra-organizational relevance, most respondents named
political issues. The instrumental aspect was measured by asking for
interaction partners, i.e. for persons who can be assumed to have tried
to influence the respondent, or whom the respondent himself bad
tried to influence. Reputation was measured in so far as respondents
were asked to name only the most important of their interaction
partners. The number of designated persons was therefore
presumably much smaller than the actual number of interaction
'
partners.
With respect to the traditional approaches of elite identification,
the nominations can be classified as a variety of the decisional
approach based on reputational nominations. The instrumental as
weil as the reputational nature of these nominations ensures that they
are not restricted solely to persons with formal power but that they
also cover informal influence relations. The approach allows, in
other words, supplementation of the original positional sample by
persons with informal power. At the same time, persons holding only
formal but no real decision-making power can be detected.
Network analytic procedures were then used to analyse these
sociometric data. 4 They allow the specification of the boundaries of
elite circles which are defined as aggregations ofhighly overlapping,
'face-to-face' cliques. Additionally, centrality measures can be
calculated for each member of the elite sample that are based on the
number and type of persons to which the sample member is linked.
They can, in turn, be used as a weight for the political influence of a
person.
The analysis revealed in all three countries the existence of a
relatively broad central circle that included members as weil as nonmembers ofthe positional elite sample from all sectors and active on
different issues. Its sector composition differed considerably from
that of the original positional sample.
Table 2.8 allows comparison ofthe sector composition ofthe West
German positional elite sample with that of the elite network and the
central circle. The network includes all respondents who nominated
295
155
354
179
172
180
218
8.0
7.0
8.2
6.5
0.5
1.1
5.1
100.0
78
69
80
64
5
11
50
979
100.0 1230 100.1
98
88
107
96
6
12
73
8.0
7.2
8.7
7.8
0.5
1.0
5.9
14.6
14.0
4.4
33.0
17.7
12.9
143
137
43
323
173
126
14.0
13.7
4.8
network
n
%
32.8
17.0
11.2
403
209
138
172
168
59
Network
%
n
37.4
13.6
11.8
7.3
7.9
8.2
8.2
0.2
0.5
4.8
209
76
66
41
44
46
46
1
3
27
31.9
14.3
4.8
8.0
7.6
10.8
12.7
0.4
0.4
9.2
100.l
80
36
12
20
19
27
32
1
l
23
251
559
16.l
15.0
6.1
90
84
34
11.6
12.4
6.4
29
31
16
99.9
%
n
circ/e
Central
network
%
n
Other
personsin
495
bIncluding foreigners nominated
38
38
39
37
1
3
22
185
68
64
83
73
29
100.0
7.7
7.7
7.9
7.5
0.2
0.6
4.4
37.4
13.7
12.9
16.8
14.7
5.9
centra/ circ/e
%
n
Sample
membersin
Notes:
a lncluding non-sample politicians with other party affiliation or whose party affiliation could not be ascertained
3164
9.3
4.9
11.2
5.7
5.4
5.7
6.9
452
471
688
Politics totala
Civil service
Business
Business
associations
Trade unions
Massmedia
Academic
Military
Culture
Otherb
Total
14.3
14.9
21.7
199
208
45
Politics SPD
CDU/CSU
FDP
6.3
6.6
1.4
Samp/e
%
n
membersin
Sample
Seetor composition of sample, network, and central circle
Seetor
Table2.8
64
5
3
6
7
9
0
0
100.0
4.7
9.4
10.9
14.l
0.0
0.0
7.8
37.5
12.5
3.1
24
8
2
5
%
10.9
17.2
7.8
circle
7
ll
n
incentral
Other persons
44
RESEARCH METHODS FOR ELITE STUDIES
other persons and were themselves nominate? by at least on~ other
respondent. Non-respondents were included 1fthey had rece1ved at
least two nominations.
The results show that only about one-third ofthe members ofthe
original positional elite sample belong to the network and only 15.6
per cent to the central circle. At the same time, however, only 2? 1
persons who were not holders of elite positions entered the ehte
network, and even fewer of these, the central circle. Positional power
and political influence as measured by the sociometric approach
have therefore tobe conceived as different though related concepts.
2.6 Conclusion
Compared to a general population survey, the collection of survey
data on national elites poses a number of additional problems. The
organization of the field work requires more effort with regard to
getting interview appointments, adequate training of interviewers,
and data protection. As a number ofnational elite surveys-not only
in West Germany but also in other countries- have shown, the use of
a highly structured questionnaire presents no obstacle and does not
lead to insufficient response rates. The decision to use a structured
interview guide has to depend solely on considerations ofthe research
goal and not on imputed reservations of respondents to forced choice
questions.
The validity of the results depends to a large extent on the sampling
design. The positional approach is widely used for reasons of
practicability. In most cases, however, the researcher does not just
want to study a sample of position-holders in different sectors, but
also wants to generalize the results to 'the elite', i.e. the group of the
most influentialpersons in a society.
The positional approach precludes such inferences for several
reasons. The first is that political influence and position are only
imperfectly correlated. Secondly, due to the multidimensionality of
power resources, power is not comparable across sectors. A uniform
boundary criterion is needed in order to make cross-sectoral
c~mparisons. The same is true for determining the overall size of an
ehte. Normally, the composition of the positional elite sample in
tern;is of sectors, organizations, and positions, is used as a weight for
the 1mportance ofthe different subgroups. Varying degrees ofpower
concentration and multiple position-holding, however, may
preclude the realization of this intention and varying response rates
may additionally distort the intended numerical relations. Finally,
NATIONAL El/TES IN GERMANY
45
unequal power within the elite precludes inferences from distributions of attitudes within the sample of respondents to future
decision-making outputs.
In order to identify the politically influential among the members
of a positional elite sample as weil as among persons not holding top
positions, we need, therefore, an additional empirical indicator of
political influence. This should allow the determination of the
boundaries of an elite and should at the same time provide a
quantitative measure that can be used as a weight for individual
respondents.
The design used in the United States, Australian and West German
elite studies tries to make up for the above-mentioned shortcomings
of the positional approach. Starting out from a positional sample,
respondents were asked to name other persons relevant to decisionmaking in their own field of activity. A network analysis of these
nominations allows the detection of the network of interactions
among elites and the central circle ofthis network. lt also provides a
measure of the centrality of persons in the network of relevant
decision-makers, a measure that can be conceived as a weight of
political influence. This approach, therefore, enlarges substantially
the evidence that can be obtained from national elite surveys.
Appendix: The sample of the West German elite study, 1981: sectors,
organizations, and positions
Politics
- Federal govemment: chancellor, ministers, andjunior ministers;
- State govemments: prime ministers, ministers, andjunior ministers;
- Federal legislature ('Bundestag'): president, vice~presidents, chairmen
and deputy chairmen of the standing committees; leaders of the
parliamentary parties;
- State legislatures: leaders ofthe parliamentary parties;
- Political parties: members of the national committees; chairmen and
deputy chairmen ofthe state committees.
2 Civil Service
- Federal ministries: secretaries of state, department heads (political civil
servants) 3 ; subdepartment heads (permanent civil servants);
- State ministries: secretariesofstate (political civil servants);
- Federal and state agencies: directors, deputy directors.
3 Business
- Industrial, trade, and service corporations according to size of sales: chief
executives, chairmen and deputy chairmen ofthe supervisory boards;
- Financial corporations (banks, insurances) according to size of sales:
46
4
5
6
7
8
9
10
-
RESEARCH METHODS FOR EUTE STUDIES
chief executives, chairmen and deputy chairrnen of the supervisory
boards;
Federal bailk: members ofthe executive board ('Zentralbankrat').
BusinessAssociations
Peak associations of industry and employers: boards of directors, chief
executives;
Agricultural associations: presidents, vice-presidents, chiefexecutives.
TradeUnions
German trade union federation (Deutscher Gewerkschaftsbund, 008)
and its member unions: members ofexecutive boards, district heads;
Union of employees (Deutsche Angestelltengewerkschaft, DAG):
members ofexecutive board, department heads.
MassMedia
Press (dailies, weeklies, and magazines) according to number of
circulation: executive managers, chief editors, chief editorial staff of
political and economic sections;
Broadcasting networks: executive managers, program directors, chief
editorial statfof political and economic sections.
Academic
Universities: presidents;
Non-commercial research institutes: presidents, department heads;
Research departments of large industrial corporations: department
heads;
Public and private research foundations: presidents ·and chief
executives;
Economic advisory committee to the federal government: all members.
Military
West German armed forces ('Bundeswehr'): all generals and admirals
including those in the NATO staff.
Cultural
Press and broadcasting networks: chief editorial statf of cultural and
entertainment sections;
Publishingcompanies: directors, chiefexecutives, and editors.
Other
Judiciary: presidents and chairmen ('Senatsvorsitzende') of all federal
courts including the federal constitutional court;
Local elites: mayors and administrative heads ofthe biggest cities;
Churches: protestant and catholic bishops;
·
Professional associations: presidents and managing directors of the
associations ofthe medical, legal, and cultural professions as weil as the
civil servants' association (Deutscher Beamtenbund);
Consumers' associations ('Arbeitsgemeinschaft der Verbraucher'):
presidents, vice-presidents, managingdirectors.
Note:
~ Political ~ivil servants can be removed from their positions and sent into temporary
ret1rement without further explanation.
NATIONAL ELITES IN GERMANY 47
Notes
We have tried to make up for this shortcoming by including some open-ended
questions on the most important problems which the Federal Republic is facing
today.
2 The study was carried out by a research team at the University of Mannheim.
Prindpal investigators were Rudolf Wildenrnann, Max Kaase and the author. lt
was supported by a grant of the Deutsche Forschungsgemeinschaft (DFG). The
major part of the field work was organized by GET AS, Bremen. ZUMA,
Mannheim, provided assistance during all stages ofthe project and particularly in
the preparation ofthe data sets. The Zentralarchiv für empirische Sozialforschung,
Cologne, produced a machine-readable code-book containing the marginal
distributions ofthe answers for the different elite subgroups (sectors): authors were
RudolfWildenmann, Max Kaase, Ursula Hoffmann-Lange, Albrecht Kutteroff,
Gunter Wolf, Führungsschicht in der Bundesrepublik Deutschland 1981.
Mannheim: Universität Mannheim, 1982.
3 See Appendix for a !ist ofthe sectors and positions included in the study.
4 The programs SOCK and COMPLT developed by Richard D. Alba were used for
this purpose.